多分组差异分析解决方案（2）分批次差异基因后取交集

#===========================================================================


#===========================================================================


rm(list = ls(all.names = TRUE))

options(stringsAsFactors = F)

library(Matrix)



setwd('D:\\SCIwork\\F38KRT\\s2')


data <- read.csv('cdata.csv', header = T, row.names = 1)

data <- as.data.frame(t(data))

data[1:4,1:4]


normalized<-function(y) {
  
  x<-y[!is.na(y)]
  
  x<-(x - min(x)) / (max(x) - min(x))
  
  y[!is.na(y)]<-x
  
  return(y)
}

db <-  as.data.frame(apply(data,2,normalized))


data <- db

data$sample <- rownames(data)

data$sample <- chartr(old='.', new='-', x=data$sample)


setwd('D:\\SCIwork\\F38KRT\\s3')

group <- read.csv('group2.csv', header = T)

names(group)[1] <- 'sample'

group$sample <- chartr(old='.', new='-', x=group$sample)

group <- subset(group, select=c("sample", "group"))

group$subtype <- group$group

group$group  <- NULL




dt <- merge(group, data, by='sample')

dt[1:4,1:4]

dt$sample <- NULL

table(dt$subtype)

dt_total <- dt





normalized<-function(y) {
  x<-y[!is.na(y)]
  x<-(x - min(x)) / (max(x) - min(x))
  y[!is.na(y)]<-x
  return(y)}

#===========================================================================


#===========================================================================


table(dt$subtype)

dt <- dt[order(dt$subtype), ]

dt[1:4,1:4]



dt_Con <- subset(dt, dt$subtype == 'Subtype1')

dt_Con[1:4,1:4]

dt_Exp <- subset(dt, dt$subtype == 'Subtype2')

dt_Exp[1:4,1:4]




dt_Con$subtype <- paste0(dt_Con$subtype, rownames(dt_Con))

rownames(dt_Con) <- dt_Con$subtype

dt_Con$subtype <- NULL

dt_Con <- as.data.frame(t(dt_Con))




dt_Exp$subtype <- paste0(dt_Exp$subtype, rownames(dt_Exp))

rownames(dt_Exp) <- dt_Exp$subtype

dt_Exp$subtype <- NULL

dt_Exp <- as.data.frame(t(dt_Exp))


Pvalue<-c(rep(0,nrow(dt_Con)))

log2_FC<-c(rep(0,nrow(dt_Con)))

for(i in 1:nrow(dt_Con)){
  
  y=t.test(as.numeric(dt_Con[i,]),as.numeric(dt_Exp[i,]))
  Pvalue[i] <- y$p.value
  log2_FC[i] <-log2(mean(as.numeric(dt_Exp[i,]))/(mean(as.numeric(dt_Con[i,]))))
  
}


library(dplyr)

library(tidyr)

library(tibble)

# 对p value进行FDR校正
fdr=p.adjust(Pvalue, "BH") 
# 在原文件后面加入log2FC，p value和FDR,共3列；
out<- as.data.frame(cbind(log2_FC,Pvalue,fdr))
out$gene <- rownames(dt_Con)
# out <- out %>%
#   dplyr::filter(log2_FC > 0.5 & Pvalue < 0.05)


setwd('D:\\SCIwork\\F38KRT\\s5')

write.csv(out, file = 'out_S1.csv')

#===========================================================================


#===========================================================================


table(dt$subtype)

dt <- dt[order(dt$subtype), ]

dt[1:4,1:4]



dt_Con <- subset(dt, dt$subtype == 'Subtype2')

dt_Con[1:4,1:4]

dt_Exp <- subset(dt, dt$subtype == 'Subtype3')

dt_Exp[1:4,1:4]




dt_Con$subtype <- paste0(dt_Con$subtype, rownames(dt_Con))

rownames(dt_Con) <- dt_Con$subtype

dt_Con$subtype <- NULL

dt_Con <- as.data.frame(t(dt_Con))




dt_Exp$subtype <- paste0(dt_Exp$subtype, rownames(dt_Exp))

rownames(dt_Exp) <- dt_Exp$subtype

dt_Exp$subtype <- NULL

dt_Exp <- as.data.frame(t(dt_Exp))


Pvalue<-c(rep(0,nrow(dt_Con)))

log2_FC<-c(rep(0,nrow(dt_Con)))

for(i in 1:nrow(dt_Con)){
  
  y=t.test(as.numeric(dt_Con[i,]),as.numeric(dt_Exp[i,]))
  Pvalue[i] <- y$p.value
  log2_FC[i] <-log2(mean(as.numeric(dt_Exp[i,]))/(mean(as.numeric(dt_Con[i,]))))
  
}


library(dplyr)

library(tidyr)

library(tibble)

# 对p value进行FDR校正
fdr=p.adjust(Pvalue, "BH") 
# 在原文件后面加入log2FC，p value和FDR,共3列；
out<- as.data.frame(cbind(log2_FC,Pvalue,fdr))
out$gene <- rownames(dt_Con)
# out <- out %>%
#   dplyr::filter(log2_FC > 0.5 & Pvalue < 0.05)


setwd('D:\\SCIwork\\F38KRT\\s5')

write.csv(out, file = 'out_S2.csv')

#===========================================================================


#===========================================================================


table(dt$subtype)

dt <- dt[order(dt$subtype), ]

dt[1:4,1:4]



dt_Con <- subset(dt, dt$subtype == 'Subtype1')

dt_Con[1:4,1:4]

dt_Exp <- subset(dt, dt$subtype == 'Subtype3')

dt_Exp[1:4,1:4]




dt_Con$subtype <- paste0(dt_Con$subtype, rownames(dt_Con))

rownames(dt_Con) <- dt_Con$subtype

dt_Con$subtype <- NULL

dt_Con <- as.data.frame(t(dt_Con))




dt_Exp$subtype <- paste0(dt_Exp$subtype, rownames(dt_Exp))

rownames(dt_Exp) <- dt_Exp$subtype

dt_Exp$subtype <- NULL

dt_Exp <- as.data.frame(t(dt_Exp))


Pvalue<-c(rep(0,nrow(dt_Con)))

log2_FC<-c(rep(0,nrow(dt_Con)))

for(i in 1:nrow(dt_Con)){
  
  y=t.test(as.numeric(dt_Con[i,]),as.numeric(dt_Exp[i,]))
  Pvalue[i] <- y$p.value
  log2_FC[i] <-log2(mean(as.numeric(dt_Exp[i,]))/(mean(as.numeric(dt_Con[i,]))))
  
}


library(dplyr)

library(tidyr)

library(tibble)

# 对p value进行FDR校正
fdr=p.adjust(Pvalue, "BH") 
# 在原文件后面加入log2FC，p value和FDR,共3列；
out<- as.data.frame(cbind(log2_FC,Pvalue,fdr))
out$gene <- rownames(dt_Con)
# out <- out %>%
#   dplyr::filter(log2_FC > 0.5 & Pvalue < 0.05)


setwd('D:\\SCIwork\\F38KRT\\s5')

write.csv(out, file = 'out_S3.csv')

#===========================================================================


#===========================================================================


diff1 <- read.csv('out_S1.csv', header = T, row.names = 1)

diff2 <- read.csv('out_S2.csv', header = T, row.names = 1)

diff3 <- read.csv('out_S3.csv', header = T, row.names = 1)



diff1on <- subset(diff1, diff1$log2_FC > 0.2)
diff2on <- subset(diff2, diff2$log2_FC > 0.2)
diff3on <- subset(diff3, diff3$log2_FC > 0.2)

up_gene <- intersect(diff1on$gene, diff2on$gene)
up_gene <- intersect(up_gene, diff3on$gene)




diff1down <- subset(diff1, diff1$log2_FC < -0.2)
diff2down <- subset(diff2, diff2$log2_FC < -0.2)
diff3down <- subset(diff3, diff3$log2_FC < -0.2)

down_gene <- intersect(diff1down$gene, diff2down$gene)
down_gene <- intersect(down_gene, diff3down$gene)